Identifying Phonological Barriers in Korean-Accented English Using Artificial Intelligence Apps

Foreseeing the future of English education, this study has incorporated artificial intelligence voice assistant mobile phone applications (AI apps) into English Language Teaching. The study explored: (1) to what extent the current AI apps can dictate Korean-accented English (KoE), and (2) the phonological elements that are commonly misrecognized by the current AI apps. To accomplish these goals, 108 Korean undergraduates, whose English proficiency levels were on average B2 (upper-intermediate level) according to the Common European Framework of Reference for languages (CEFR), participated in the study. Each participant read twenty-five sentences aloud, while AI apps produced transcripts of their dictation. Then the percentage of correctly recognized target words was calculated to determine speech intelligibility. The findings show that KoE attained an accuracy rate of 56.8 percent on average, ranging from 50.1 to 62.7 percent. The KoE barriers associated with the phonological features of less intelligible tokens were also investigated. By bringing attention to the features in comparison to the Lingua Franca Core targets (Jenkins, 2000), the findings suggest the necessity of further research into the feasibility of implementing AI apps within teaching practices, and expanding the boundaries of English as a lingua franca research to KoE, which has not been sufficiently studied.